nTrain_, nTest_ randomly picked in each process sample for N>1 background processes?

Hello,

say that in a binary classifcation I am adding trees for different processes (i.e., describing different physics processes) to the background sample:


dataloader.AddSignalTree(treeS, 1.0)
dataloader.AddBackgroundTree(treeBkg1, 1.0)
dataloader.AddBackgroundTree(treeBkg2, 1.0)
dataloader.AddBackgroundTree(treeBkg3, 1.0)

If I set theSplitMode=Random option, and set a number for nTrain(Test)_Signal and nTrain(Test)_Background, will it be ensured that the subsample of training(testing) events will be chosen randomly among the various background processes trees? Or is there anything else I should take care of?

Thanks!

Marco

Hi,

Using nTrain(Test)_Signal(Background) with SplitMode=Random will sample from the signal and background classes uniformly, yes. You don’t have to do anything else :slight_smile:

Cheers,
Kim

Ahh, good to hear that :slight_smile:

Cheers,

Marco